That one horrendous in-house legacy application that should be ditched, but no one is brave enough to start building replacement from scratch ... Every company has one.

And it's written in COBOL and is part of the core process.

@vringar Not necessarily - it might be mess of various modules built on different platforms, languages,... with duct-tape to hold everything together. Years of development, adding functionality, layer upon layer ... And, yet, it works (somehow). ;)

I think you have to be very very confident to be sure that you can rewrite correctly something that was developed for many years, by many different developers and works on production for a long time. I dont say that is not possible but in most cases is much harder that it seems.

@Nekon no, no, no... mere confidence isn't sufficient. you'll need to spend time, money, dedicated team who understand all those legacy function and know what to do to make better replacement, a clone of current system infrastructure for development, etc etc etc

@Nekon Of course, but, after that many years of development, too many people have left project, new people come, so old code begins to act as some kind of magic black box that produces results and nobody actually knows what is going on under the hood.

Although there are too many problems involved with starting from scratch sometimes it might be reasonable (even if a bit pricey).